Sharkmon - deep packet monitoring 

 

  • dive deep into the contents of thousands of PCAP trace files in a single dashboard.
  • using Shark syntax for enabling more than 100,000 protocolfields
  • Masses of data easy to organize, aggregate, analyze and prioritize.
  • Grouped into 3 main categories - application - connection or network
  • allows quick assignment of errors and incidents
  • data saved in database - long history

why sharkMon

IT Data is network data - network packets travel the whole IT delivery chain - and transport the information about status and performance between endpoints: DNS and LDAP codes and times, Network and Application Performance, Server Responsetimes & Return Codes, Frontend / Backend Performance - or any content of a readable packet.

If Application is slow, Service not reachable, Error codes - infos about these symptoms - and often about the causes - can be found in packets.

sharkMon

- imports network packet data - and does provide required performance and status metrics conained in such packets.

- is using pcap files generated everywhere in the network - in the cloud on servers, at user PCs, firewalls, capture appliances etc. and can aggregate distributed capture sources into a single monitoring pane.

It can create incidents for root cause analysis and incident correlation and forward those incidents into central event correlation systems.

 

sharkMon - at a glance

- longtime data - import realtime large numbers of pcap files for hours, days, weeks - created by various trace tools like Tcpdump, Tshark, or a capture appliance
- Auto-Analysis - analyze thousands of sequential files automatically on the fly by using customizable deep packet expert profiles - also per object - including custom metrics and thresholds
- incidents - create incidents based on variable thresholds per object
- longtime perspective - visualize incidents and raw data in smart dashboards over hours, days, weeks or months
- Incident correlation - Export incidents into service management management, becoming part of the correlation framework
- automation - Automate the analysis workflows step by step - avoiding time and efforts for recurring tasks

Longtime monitoring - or single tracefile analysis?

Trace files are usually manually analyzed in single steps - just one at a time, covering a few minutes. For hours multiple files must be generated - for a day 100 or 1000 of files. This cannot be done manually. with sharkMon user can import a large number of files from servers, cloud or datacenter appliances - assign an analysis profiles including the relevant metrics - and sharkMon creates required statistics and over the whole span of time - just monitoring.

smart dashboards

Just with a glance a user can understand:

  • Are there any issues in my trace files
  • To what category they belong too (network, application, connection)
  • Which exact metric was causing that?
  • What threshold was crossed
  • Direct access to the trace file
  • Drilldowns and category specific
  • views (here application view) allow deep insights - continuously over time - for days, hours or seconds

Deep analysis

With Deep analysis InterTrace is utilizing Wireshark display filters - which can do a lot more than most other analysis solutions. Thousands of protocol-dependent prefilters are defined, analysis expert exist for a wide range of protocols. By using each possible Wireshark-Display filter in sharkMon - user can pretty much use every byte in the packet flow - as monitoring and incident condition.

sharkMon under the Hood

Analysis Profiles

Analysis profiles are pre-configured customizable filter and threshold definitions which will be applied to a trace analysis. A profile is a configuration of defined filters and symptoms - pretty much each byte in a packet or a Wireshark-expert-analysis (like tcp_out_of_order) can be configured as symptom. Files will be analyzed very deeply according to these profiles - and symptoms are generated based on the analysis. Eg. if SSL uses TLS1.2 can be defined as condition, an occurrence on non-TLS1.2 packets can be seen and defined as symptom. Same can be done with performance metrics like LDAP.time, DNS.Time, DNS. response codes, HTTP return codes etc. - which can be included in a specific profile and symptoms created if a threshold is exceeded.

The front-end and back-end server systems are listed or shown in an architecture chart.
Since the service discovery is carried out daily, the architecture charts are always up-to-date.
Changes within a service chain, e.g. new servers, are recorded and reported.

Scenarios

User define their analysis objects and metrics as an analysis scenario:

  • Object - What I need to analyze
  • Conditions - filter conditions, time (backward or future)
  • Data source - PCAP files, active Wireshark / tcpdump srtewam, capture appliance
  • Options - for analysis purpose (like de-duplication, merging).
  • Intelligence - What analysis profile should be used, which includes metrics thresholds etc.¬†

Such a scenario gives the user the ability to start a longtime-monitoring process on a deepest level - focus on this scenario and create scenario-related incidents and events. Many scenarios can be defined and processed parallel - so one scenario can work on the web shop using deep SSL and HTTP metrics, another can monitor SAP services and another the DNS replies - same time.

Correlation

Trace-based events can be correlated with other existing management data, if coming from network, systems, log files or security devices in a single dashboard-like SLIC Correlation insight. They can create the significant data - which can feed a service management platform with the intelligence to create complete cause & effect chains for complex IT services.